4
numerical simulation. Therefore, it can reduce the optimization time significantly.
Different DOE methods and surrogate modeling methods for approximate models are
studied and compared. For the optimization of high-speed multi-station forging process, the
combination of LHS and Kriging is proposed for the case of numerical response value, and BP
network is used for the case of linguistic response value. Two DOE methods, which include
orthogonal experimental design and LHS, together with four modeling methods, which include
least square (LS) response surface, moving least square (MLS) response surface, BP neural
network and Kriging model, are applied to establish the approximate relationship models
between variables and response values of an actual case. And the prediction accuracies of the
different models are compared.
In order to automatically obtain the response values corresponding to the modification of
design, the CAD/CAE intelligent integration technique is proposed in this thesis. It provides an
intelligent integration platform for optimization to reduce the complex interactive operation. It
contains the processes of the automatic modification of tools’ geometry models according to
variables, the intelligent modeling of CAE analysis model and the automatic acquisition and
feedback of the CAE simulation result. The knowledge integration method based on template of
parameterized geometry model and optimization case model is proposed. Through the
optimization case model, variables are corresponding to the parameters of tools’ geometry
models in one or different forming stations automatically, and the highly integration of
optimization information and geometry information is realized. Based on the parameterized
geometry template of cavity, the automatic acquisition technique is applied to obtain the tools’
geometry models according to the values of variables, which ensure the correct constraints and
assembly relationships between tools. The knowledge based CAE intelligent modeling technique
is proposed. It achieves the intelligent establishment and automatic modification of CAE
analysis model, so that the simulation of the forging process is carried out by sequence of
stations automatically. The knowledge based intelligent post simulation feedback technology is
proposed. Based on knowledge acquisition and intelligent model reconstruction, the useful
information is analyzed and transformed from the CAE simulation result. Therefore, the
automatic mapping process from simulation result to response values (objective functions and
constraints) is implemented. Further more, the intelligent integration and automatic process,
feedback of response values according to the modification of variables, is implemented.
Based on the studies of surrogate model based combined optimization strategy and
CAD/CAE intelligent integration technique, an intelligent optimization system for high-speed
multi-station forging process is developed on the platform of UG NX and DEFORM by using
the software of Visual C++.net, MATLAB and EXCEL. The surrogate model based optimization
method and related knowledge is integrated in the intelligent optimization system, which guides
designers to complete the optimization task of high-speed multi-forging process. Meanwhile, the
intelligent system provides effective intelligent support in the process of optimization. The
optimization case study of combring produced by high-speed multi-station forging has
demonstrated the reliability and practicality of the intelligent optimization system.
